Saturday, June 5, 2010

WebM (VP8): a license update and the gift horse fallacy

I have reported repeatedly on the topic of "patent-free" multimedia codecs, such as in this roundup describing the WebM situation about two weeks ago. In discussions (such as on slashdot) there's always a lot of confusion about the subject and, unfortunately, about the intentions of those who issue Cassandra warnings.

License update

Since my last post on this subject, the only significant news has been that Google modified the WebM licensing terms so as to make the WebM license consistent and compatible with existing FOSS licenses. This has been welcomed by the Free Software Foundation as well as Simon Phipps, a board member of the Open Source Initiative.

However, Google hasn't changed a thing about the risk that allegations of the infringement of third-party patents (Google's own patents have never been a problem here) by WebM represents to its adopters. There still isn't any hold-harmless or indemnification clause.

We must look every gift horse in the mouth

I think there are many people in the FOSS community who would benefit from a more differentiated approach toward anything that appears to be a generous donation. At the end of the day, the Trojan Horse was a gift horse, too (just to avoid any misunderstanding: I don't mean to liken WebM to the Trojan Horse; this was only meant to point out that refusing to look in the mouth is a dangerous approach).

The more realistic perspective -- still not a cynical one -- is that companies don't give for the sake of giving. If they give, they expect some kind of benefit in return. In some cases, the "free" deal can be very beneficial and fair. I use some Free Software that gives me great value and I can't see a risk. But at the other end of the spectrum there are "freebies" that are indeed modern variations of the Trojan Horse, such as those "open source patent pledges" by IBM and others.

When IBM first made that (in)famous pledge in early 2005, there were only a few people who understood that what seemed to be an act of generosity was a bad thing on the bottom line. Unfortunately, even an open-innovation luminary like Professor Lawrence Lessig got it wrong. He said: "This is exciting. It is IBM making good on its commitment to encourage a different kind of software development and recognizing the burden that patents can impose." Five years later, IBM proved him completely wrong on this. IBM continued to be just the same patent bully it's always been.

Google's WebM initiative is somewhere in the middle between a true act of generosity and an IBM-style scheme:
  • There's no reason to assume that Google wants to hurt the FOSS cause in any way with WebM, especially not in any IBM-like way. I don't put it past Google to have that intention elsewhere: they might do anything, including the use of patents, to destroy an open source search technology that could adversely affect their core business. However, in this particular context of video codecs, I don't think they intend to cause harm. I do believe them that they want more competition in this case.

  • What Google does do -- and what I believe the FOSS community must approach cautiously -- is to shift most of the risk to others while keeping most of the benefits to itself. Businesses like to do that, but FOSS developers and users shouldn't lose sight of the risks just out of excitement over the idea of getting a seemingly "unencumbered" codec.

Google will retain control over WebM despite open-sourcing program code and publishing specifications

A common misconception about open source and "free" specifications is that this would make something such as the WebM project independent from a single vendor or a group of vendors. Some think this puts "the community" in charge.

Theoretically, it's true because it would be legally possible. Practically, it's just wrong because we live in a dynamic world as opposed to a static one.

If any WebM code is released on FOSS terms, that doesn't guarantee that all future versions will be, too. The original copyright holder has every right to release future versions under different licenses.

Some will argue: well, if they do that, someone can still fork it and continue it as a FOSS project. That's not forbidden but it's rarely practical. If you do that, you have to compete against the steward of the original project. You have to give your fork a new name because you won't own any rights to the trademark of the original trademark. So the odds are long against you at the outset. Chances of a fork even being tried in any serious way are therefore quite limited; chances of such an effort succeeding are extremely low.

The bottom line is that Google will have all of the expertise in-house to develop WebM further, and this will give Google a lot of control over the future direction of the project from a practical point of view. That's perfectly legitimate, but it does have to be considered. Google is a powerful player and can steer WebM in whatever direction best benefits its business interests (search, Chrome, Android, YouTube, whatever else).

Some may be more comfortable with Google in charge of such a codec than with the backers of MPEG LA's H.264. I can see why many people feel that way but one could also argue that a single player as powerful as Google is a greater long-term risk than a consortium of many companies with a diversity of strategic interests. I don't want to support the former or the latter position: I just want to stress that hoping for Google to do away with MPEG LA -- if possible at all -- means to trade in one set of issues for another.

Google won't be just another WebM adopter

Some argue that by adopting WebM itself (by using it in and for Android, Chrome, YouTube etc.) Google will become a large-scale adopter and therefore has a fair share in the patent-related risk.

Unfortunately, third-party patent holders can go after whichever infringer(s) they want to. On the one hand, there's a huge reward out there for them if they succeed and force Google to pay. On the other hand, Google will defend itself vigorously when it's attacked directly. Smaller adopters of WebM may represent lower-hanging fruit to a patent holder. Typically the first ones to suffer from patent hold-up are those who are big enough to be able to pay significant amounts but not so big that fighting them would be a huge challenge. Then over time the targets will get bigger.

In WebM's case, patent holders would be particularly likely to prey on other adopters than Google initially: Google is also standing on the sidelines as some commercial adopters of Android are being approached by patent holders, such as HTC being sued by Apple and having felt forced to agree to pay royalties to Microsoft. So a patent holder might assume that Google would also leave WebM adopters in the lurch.

Another important difference between Google and large parts of the FOSS community is that Google does like software patents a lot. Just like most patent holders, they like their own patents and dislike those held by the rest of the world, which is human (but often misguided). It would be great if one day they could change their position and oppose all software patents, but so far they've been very much in favor of patents especially in connection with their search engine. So the WebM initiative shouldn't be confused for a crusade against software patents that one might want to support as a matter of principle, "for the cause".

In my opinion, no company that favors software patents can claim to be a reliable ally of FOSS. It can at best be a business partner for the community. And that's the perspective to take on WebM: is it a good deal? Is it a fair deal?

Don't count on Google's idealism or solidarity. You'd wind up disillusioned.

The argument that MPEG LA doesn't offer indemnification for third-party patent claims either

I have the utmost sympathy for those who say that Google shouldn't be held to a higher standard than MPEG LA in terms of indemnification for third-party patent infringement claims. Yes, I know why some think it's unfair. But there's always a higher hurdle for late entrants than for incumbents. Economists refer to that phenomenon as "barriers to entry". In this case, what comes with it is a significant burden of proof.

I fully agree that it would be desirable for MPEG LA to protect licensees of its patents against third-party infringement claims. Right now, a license from MPEG LA means that they won't sue you as a licensee over the patents you've licensed from them, but if anyone else has patents that read (or are claimed to read) on the same codec, then you're exposed to that risk anyway. In other words, you don't have a legal guarantee that MPEG LA is really a one-stop shop.

So some say that if MPEG LA doesn't provide a guarantee that you don't need any patents other than theirs to implement a certain codec, then Google (which unlike MPEG LA doesn't charge but makes WebM available on FOSS terms) shouldn't be required to do so either.

The part about "Google doesn't charge" was already addressed further above. Google has chosen a business model that is FOSS-based. All of us FOSS advocates like it, but we have to understand that they're doing this for commercial reasons, not for idealism. I think their responsibility shouldn't be less just because we like their business model better. Business is business. Free or non-free. Using their search engine is also free, still there are consumer rights advocates who insist on data privacy.

The late-entrant problem here is that H.264 is already in extremely widespread commercial use. Frankly, there aren't many digital video devices out there in the worldwide marketplace right now that don't implement it. That doesn't represent a guarantee that there aren't patents outside the MPEG LA pool that could be asserted against H.264, but I haven't found any indication that anyone has ever tried. Given that it's already so ubiquitous, it's hard to see why any "trolls" wouldn't have come out in the meantime.

By contrast, VP8 is only now beginning to take off big-time and I don't think it's responsible to simply discount as "FUD" MPEG LA's recent statement concerning the possible creation of a related patent pool (another way of saying that they believe some of their patents are infringed). I don't doubt that MPEG LA has strategic motivations that would make "FUD" potentially conducive to its purposes. I just oppose the approach of dismissing it as "FUD" without further analysis.

I mentioned before (in the context of control) the fact that Google's VP8 is a one-company effort. It was started by On2, which was acquired by Google earlier this year. It was only one company, and a rather small one compared to the long list of big companies supporting MPEG LA, which represent a very large part of all of the research and development that happened in recent years (even decades) in connection with multimedia technologies. That combined power simply makes them far more likely to have a "complete" patent pool (meaning that if you license something from them, you may indeed be fine for the relevant codec without much of a risk of third-party claims) than Google. This is like a whole bunch of long-standing Goliaths versus the one and relatively young David that On2 was before Google bought it (and Google itself wasn't even active in the codec field before it acquired On2).

All's well that ends well -- but what if it doesn't?

Given my background of promoting Free Software and Open Source and opposing software patents, it should go without saying that I'd be extremely happy to see a truly free and unencumbered video codec succeed.

Most people understand that this is my attitude. A very few have accused me of spreading or supporting "FUD". Again: There's no problem here if this works out well. But that doesn't mean we should turn a blind eye to the risks involved.

I mentioned in connection with IBM's patent pool the example of Professor Lessig, who was misguided by his dream of open innovation to the extent that he believed IBM's "pledge" of 500 patents meant a commitment to a good cause. I now see some tendencies on the part of various FOSS leaders to embrace WebM in a way that doesn't take into sufficient consideration the risks to which its adopters may be exposed.

I can see why: Yes, the whole idea of Free and Open Source Software is indeed going to be in real trouble if patents prevent the creation of a truly free and unencumbered codec. When I wrote that patent problems in connection with such technologies could even result in the World Wide Web Consortium having to rethink its patent-free/royalty-free standards requirement, I didn't like that notion at all (nor did I mean to ignore structural differences between today's W3C and the way proprietary standards like GSM are set). I just described a risk of how things could evolve over time.

Let's hope for the best but not forget to prevent the worst. The worst outcome here would be if various WebM adopters ended up being sued as a consequence of their undue reliance upon vague and unsubstantiated assurances made by Google. If the wider FOSS community takes most of the risk while Google would get most of the reward, then that's not fair in a scenario where things go wrong.

I continue to believe that a detailed analysis of the patent situation should be made available by Google to the public, and that holding WebM adopters harmless in the event of third-party patent infringement claims would be the only reliable way with which Google could practically eliminate the related concerns. If that isn't provided, it doesn't mean WebM is a bad thing or that it will necessarily fail. But it does suggest that we should continue to monitor the situation, and that it's reasonable to ask tough questions, even the ones we'd rather not have to ask in the first place.

No matter how much we want free, open and unencumbered standards, we must realize that the price of ignorant pigheadedness can be very high when patents are involved. Those are costly lawsuits that can have major consequences.

If after a year or two no one has brought up patent infringement assertions against WebM, then it will be much easier to recommend it to everyone. There would always be a risk, but in that case the risk may indeed be closer to the normal patent-related risk all software developers and distributors face all the time. For now, there are reasons to fear that the risk might be considerably higher.

Always think of your own risk. Google won't protect you based on the current license terms. Free Software and open source advocates won't protect you either. They have their agenda, so you should have yours.

If you're interested in reading more about the issue of patents on standards, I would like to recommend this article on another blog ("Video Networker"), covering standards in general and discussing the WebM situation in particular (under the "Skype and Google" subhead).

If you'd like to be updated on patent issues affecting free software and open source, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents.